A Bar Code Based Quality Inspection Procedure for Residential Construction

ASC Proceedings of the 41st Annual Conference
University of Cincinnati - Cincinnati, Ohio
April 6 - 9, 2005


A Bar Code Based Quality Inspection Procedure for Residential Construction

Denise D. Gravitt, Ph.D., MSE, BSCE
Indiana State University
Terre Haute, Indiana

Advances in technology that can be used to improve residential construction processes and procedures, including quality, are not always readily adopted by the industry. The purpose of this study was to develop a bar code based quality inspection process that would demonstrate the speed and accuracy with which data can be collected and used by residential construction contractors for analysis of quality improvement trends in homes. The three phase study had a checklist form developed for use during post-construction framing inspections of homes in phase one, then Code 39 bar codes were added to the checklist and a proprietary scanner program was developed to gather inspection data in phase two. Phase three was the field study designed to compare two data collection methods. The first method used the PSC PT40 scanner and the bar coded checklist. The second more traditional method used hand completed formso data collection methods and keyboard data entry. It was concluded that for this study the scanner method was more accurate and faster than the manual method in providing the data necessary to produce reports and evaluate possible trends in quality.

Key Words: Quality; Inspection Processes; Bar Coding; Residential Construction


Introduction

Advances in technology that can be used to improve residential construction processes and procedures, including quality control, are not always readily adopted by the industry. While quality is important to residential contractors, the procedures and technology being used by contractors in their quality assurance programs vary. Residential construction quality inspection procedures and processes are not standardized. In addition, residential quality inspection data gathered through manual data collection and keyboard data entry may be incomplete, inaccurate, and time consuming to collect and process. This paper will report current residential quality inspection procedures being used by contractors and new technological advances. Lastly, the paper will describe current research and the potential use of Automatic Data Capture (ADC), specifically bar coding technology, in the quality inspection processes for residential construction.

Significance

Competitiveness among construction companies warranted an investigation addressing the use of ADC for various activities including quality inspections in the construction industry (Finch, Flanagan, & Marsh, 1996). Total Quality Management (TQM) procedures may require large amounts of data to be collected to evaluate trends and patterns in quality performance. The collection of this data could be facilitated by the implementation of ADC technologies. Checklist responses of activities to be performed in a quality control program could be bar coded. Bar coding standardized checklists has the potential to help improve overall product quality for residential contractors that may not currently perform regular repetitive inspections. Improved data collection methods for quality control inspections would allow for faster identification of problem contractors, products, crews, and construction methods. Furthermore, the data analysis would summarize trends in the quality of residential construction.

Problem Statement

Residential construction quality inspection procedures and processes are not standardized. In addition, residential quality inspection data gathered through manual data collection and keyboard data entry may be incomplete, inaccurate, and time consuming to collect and process. Processes that may increase the amount of accurate and timely quality inspection data for residential construction companies were needed.

Statement of Hypothesis

The null hypothesis for this study was that there was no statistically significant difference in the number of errors produced by manual quality inspection data collection and keyboard data entry compared to the number of errors generated in a bar code based data collection process.

Current Inspection Processes

The improvement of residential quality procedures is a current area of study. Rogers and Christofferson (1997) stated that residential construction companies must improve their organizations in order to operate more efficiently and competitively. One component of a residential contractor’s overall quality procedures, quality inspections checklists, was investigated by the author. A review of literature found that published home inspection texts are generally written by, and for, the professional home inspector not residential contractors (Becker, 1993; Cauldwell, 2001; Hoffman, 1985; Howard, 1987; Irwin, 1995; McNeill, 1979; Traister, 1997). The National Association of Home Builders (NAHB) contracted with Rogers and Christofferson to write a production manual to assist home builders improve quality as well as their processes and construction methods efficiency. The manual, published in 2002 titled Building Quality: An Operations Manual for Home Builders, also contains inspection checklists for each stage of construction that can be customized for any residential company to assist them in evaluating or measuring quality. Quality measurement is a tool to evaluate the performance of the construction contractor, measuring customer satisfaction or conformance to design and code requirements (Torbica & Stroh, 1999).

Inspection checklists, whether customized for each company or developed by other sources, are consistent in residential construction quality programs and can be used to measure the quality of homes. During interviews conducted in 2003 by the author with three different residential contractors it was found that each used custom or proprietary inspection checklists. These inspection checklists are used repeatedly during home construction. Two of the contractors had their construction managers complete the forms by hand. These forms were then sent to the home offices for data entry into the main computer database. Of these two companies, one was going to implement a Palm Pilot based inspection checklist program with daily uploads to their home office computer. There were no plans to link this application to the daily schedule or to develop a web site to house the information.

The third contractor had not previously been entering any data from the regular field inspections into computer files or sharing the information with the home office. The construction managers at each community of this contractor dealt with all issues related to the inspection results. The data was not shared with other community construction managers or the home office even though the same suppliers and subcontractors were used in multiple communities. This contractor also used Palm Pilots, but use was limited to daily schedule information. The schedules were on a website updated daily with access available to all subcontractors and suppliers. Separate variances or back charge forms were issued by all three of the contractors as needed and linked to non-conforming inspection results for homes. Limitations for the traditional manual data collection method using the checklists included inaccurate data entry by separate individuals, time between inspection completion and data availability, and lack of communication between parties that may be affected by subcontractors or suppliers with poor quality performance.

Residential Contractor Technology

One of the three residential contractors interviewed by the author in 2003 was forcing suppliers and subcontractors to use its website to find current construction schedule and delivery information for the houses under construction (T. Wickstrom, personal communication, July 15, 2003). One of the other two residential contractors said they did not wish to force their subcontractors, partners, and suppliers to use this technology. This contractor faxes each of its subcontractors, partners and suppliers updated schedules each morning and does not provide this information on a website for those who would wish to use the technology. This company currently uses a computer version of its inspection checklist. The Project Managers link their Palm Pilots each night to upload the information, but they do not have wireless connections to the home office computer database so the upload over regular phone lines takes significant amounts of time (B. Anderson, personal communication, July, 11, 2003). Wireless systems for daily upload of reports and inspections, while in development and often not perfect in operation, could expedite the process. Increased accuracy from field reports that are not processed by office clerks are just one foreseen benefit of a wireless type of system (Sawyer, 2003).

Mobile computers and punch list software programs are additional technologies that can be used in residential quality inspection processes. Rugged mobile computing terminals suitable for use on construction sites make plans, schedules, and quality inspection checklists available to project managers (Albright, 2001). Add wireless communication ability to these mobile computing terminals and quality inspections can be expedited and results quickly shared with multiple parties. Similarly, punch list software programs can be used by residential contractors to track quality issues and generate punch lists (Staff, 2002). Data can be transferred electronically to the home office and sorted for relevant information on subcontractors, items of nonconformance or any other pertinent issue.

One application of current technology not observed in residential construction, either in practice or in published literature, is the use of Automatic Data Capture, or more specifically bar coding technology. The author has explored the use of bar coded quality inspection checklists in residential applications.

Current Research

TQM programs, including quality inspections, require significant amounts of data to be collected to ensure continuous improvement. Residential contractors are performing quality inspections on their homes, but the methods for collection and processing of the data are not standard throughout the industry as demonstrated by the residential contractors interviewed in 2003. Current residential industry inspection methods using manual data collection and computer keyboard entry are often inaccurate, untimely, and prohibit sharing of data. A possible improvement to this traditional process is the incorporation of Automatic Data Capture (ADC) technology into the process.

ADC and identification technologies are currently used in industrial applications. There are many options to choose from including data buttons, RFID tags and bar codes. Fales (1990) stated that bar codes have been found to reduce costs for material and equipment tracking as well as improve quality performance due to the availability of more accurate and timely data for analysis. Bar coding, while only one of the ADC technologies available for use, is the technology of choice for many applications due to its relative ease of use and reasonable cost. Commercial construction applications have been studied in the United States since the late 1980’s by the Construction Industry Institute among others, but little documented progress has been made toward use of ADC in the industry. The author proposed the use of bar coding technology in quality inspection procedures to reduce data entry errors, speed up data collection on inspections, and expedite the sharing and evaluation of the collected data. The research was completed during the summer of 2004.


Methodology

The proposed study had three phases. During the first phase of the study, a general post-construction residential quality inspection checklist form for house framing was developed. This was done by using qualitative research methods for the creation and validation of the checklist form. A review of existing literature sources and texts was performed to establish the existence of standard residential inspection checklist forms and procedures. Items of commonality from various checklists were combined into a general post-construction residential quality inspection form for the framing inspection stage of construction. The items selected for inclusion in the checklist reflected common items considered important for quality control by residential contractors. A panel of experts was needed for the checklist validation. The criterion for expert status was deemed to be contractors with at least 20 years of residential construction experience and familiarity with residential construction quality inspections. Three experts that met the criteria established were consulted. In addition to the minimum criteria, they also each owned a construction company and were licensed in at least one state. They were each given instructions to review the form layout and questions for clarity, accuracy, and key components required for a quality review of house framing work. The experts suggested minor corrections or language clarifications which were shared with the other expert reviewers in order to reach a consensus. The final version of the checklist was used in phase two.

The second phase of the study included the development of a bar coded version of the validated checklist form, creation of a database and the bar code scanner program. Figure 1 in Appendix A shows a partial bar coded framing checklist form. The scanner program was written using the proprietary software provided by the scanner manufacturer. The PSC PT40 Falcon was selected for use in the study due to its small size, low cost, large key pad buttons and ease of use. Figure 2 in Appendix A shows one component of the custom bar code scanner program developed for this study with the provided PSC software. Figure 3 of Appendix B shows the scanner unit and instructions on how to perform and save the framing inspection checklist data for each of the 28 home inspections.

Phase two also included the validation of the developed inspection process by this researcher in a controlled environment. This was done by entering predetermined mock inspection results via the bar code scanner into a computer database. Reports were generated to ensure the process and all computer programs were operational and accurate.

Phase three of the study included the field study for validation of the process in field conditions. The construction company selected to conduct the field study was a residential contractor with multiple divisions located throughout the United States and had over 24,000 national home starts in 2003. The company, a community developer with a branch office in Indianapolis, IN, had over 550 home closings in 2003. The homes that were used for the field study were selected from a cross section of homes under construction in the multistage cluster sample of communities in the Indianapolis, IN by the selected company. The field study consisted of three inspectors gathering data from 28 separate inspections. These inspectors were trained in the proper procedures for conducting and saving the inspection data using this study’s two methods. The three inspectors all used the framing checklist form developed by this researcher. One form was completed by hand and was used as the control or Master inspection for each house. The second and third framing checklist forms were completed by two different data recording methods. One of the other two inspectors completed the standard inspection form by checking the responses manually and then later entering the data via a computer keyboard into an Excel spreadsheet for the manual method inspections. The third inspector performed the same inspection but used the bar coded checklist form and a PSC PT40 bar code scanner to read and save the appropriate inspection data and then exported the data into an identical Excel spreadsheet. Table 1 below shows a partial section of the Excel file for the manual method inspection data. Table 2 shows a portion of the data file stored in the PT40 scanner collected from the checklist forms. The comma delimited format allowed the data to be linked into an Excel file matching the format for the Manual Method Checklist Data as shown in Table 1. The data from the completed inspections were compared to the control or Master inspection checklist forms for an evaluation of accuracy in the manually completed and scanned inspection checklist data. The data relating to numbers of inaccurate entries for each method were analyzed with nonparametric statistics and evaluated for any differences.

Table 1 Partial Keyboard Entered Manual Method Checklist Data
Insp. Time	Insp. Date	Community Name	Lot #	Project Mgr	Subcontractor	Question Answer	Comment Form #	Var. #
1:03:00 AM	8/16/2004	Geist	12	Smith	a-1	301yes
1:03:00 AM	8/16/2004	Geist	12	Smith	a-1	302yes
1:03:00 AM	8/16/2004	Geist	12	Smith	a-1	303no	1

Table 2
Partial Uploaded PT 40 Text File

Time, Date, Community, Lot #, Project Mgr., Subcontractor, Question Answer, Comment #, Variance #

14:28:13,08/20/04,*GEIST*,12,*SMITH*,*A1 FRAMING*,*301YES*,,

14:28:13,08/20/04,*GEIST*,12,*SMITH*,*A1 FRAMING*,*302YES*,,

14:28:13,08/20/04,*GEIST*,12,*SMITH*,*A1 FRAMING*,*303 NO*,CF001,


The data collection occurred over two days with the selected contractor employees. The control or Master checklists were completed by the Regional Construction Manager who indicated the appropriate responses of the inspection checklist questions and verbally conveyed these answers to the other two inspectors for each of the 28 house inspections. The manual process and the bar code scanner process inspectors were selected by the Regional Construction Manager conducting the field study. They were both Project Managers for the company with Construction Management degrees earned within the last two years and had comparable computer skills. After the completion of the inspections and marking the checklist forms to indicate the desired question responses, data entry for the manually completed checklists and the scanning of the bar coded forms began. The typed data entry of the manually completed checklists begun on August 16 was completed on August 20, 2004. All of the scanned data from the checklists was collected on August 20, 2004. Once complete, the data was uploaded into a text file created for the scanned data, and then imported into an Excel spreadsheet. This spreadsheet, and the spreadsheet containing the manually completed checklist data entered via computer keyboard, was linked to the Access database created for detailed summary report generation to facilitate error analysis.


Results

The checklist summary reports for the two methods were used to evaluate the accuracy of the data entered into the database through comparison with the Master inspection checklists. There sum total of all entries for both methods was 2014 based upon the 28 inspection checklists; each method had 34 data entries for each checklist plus 45 Comment Form numbers and ten Variance numbers. Out of the 1007 data entries for the manually completed checklists there were nine errors in them due to a failure in communication between the Master inspector and the person completing the manual inspection checklists. The wrong response was marked on the checklist forms. Similarly, in the 1007 data entries for the bar code based method there were three communication errors. If the communication error resulting from incorrect question responses marked on the forms were typed or scanned correctly into the databases, they were not included in the error totals for either method.

The manual checklist method with keyboard data entry resulted in 78 data entry errors and 324 copy or Excel auto fill errors for a total of 402. The 78 manual data entry errors included incorrect spelling, numbers not typed in correctly, missed question numbers, missed comment form numbers, missed variance numbers, missed lot number changes, and incomplete entered numbers or those missing digits and prefixes. Since both inspectors chosen by the company performing the field study had comparable computer skills and facility with technology, this number of errors is significant. The data entry process was conducted at a field office location where interruptions from subcontractors and phone calls occurred as they would in a real industry scenario.

The scanned data method had eight data entry errors, five due to missed question responses (they should have been scanned but were skipped), two due to the wrong response being scanned, and one missed Variance number. There were also four instances of partially scanned bar codes, resulting in a total of 12 errors for the scanner method as shown in Table 3. Since once the checklist data was scanned and the data entry process was automated the potential for interruptions during data entry was less and the process overall took much less time.

The data entry error values indicate that for this study there was a 3.87% error rate for the keyboard data entry versus a 0.397% error rate for the scanner data entry method.

Table 3 Summary of Results for Both Data Collection Methods
METHOD	DATA ENTRY ERRORS	TOTAL ERRORS	TOTAL DATA ENTRIES	DATA COLLECTION (minutes)	EXCEL FILE DATA ENTRY (minutes)
Scanner	8	12	2014	56	2
Manual	78	402	2014	28	153


The comparison of the database reports with the Master checklist data allowed the researcher to count the number of errors generated by the inspectors using the two different data collection methods. The total errors counted and the chi square analysis is shown in Table 4. For one degree of freedom, contingency tables indicate the critical chi-square value for an a of 0.001 is 10.83, or only a 0.1% chance of incorrectly rejecting the null hypothesis (Sheskin, 1997). For this study, the null hypothesis was that the error rates between the two methods were the same. Thus, it is appropriate to reject the null hypothesis for the calculated c²value equal to 462.453 per Table 4. With less than 0.1% chance of error, it can be concluded that the data entry errors that occurred during the keyboard entry of the manually completed inspection checklists were related to the manner of data entry, and the scanner method was more accurate than the manual method.

Table 4 Chi-square Test of Independence Results for All Errors
CHI-SQUARE RESULTS	y_ij	e_ij	y_ij – e_ij	(y_ij – e_ij)²	[(y_ij –e_ij)²/e_ij]
No Errors/Scanned	995	800	195	38025	47.53125
Date Entry Errors/ Scanned	12	207	-195	38025	183.6956522
No Errors/ Manual	605	800	-195	38025	47.53125
Data Entry Errors/ Manual	402	207	195	38025	183.6956522
TOTAL	c²=				462.4538043


The time expended to perform the inspections for the two different methods was of interest to the author. Since this was a new process that had never been done no hypothesis on time for the two methods was theorized. Time data was collected by the researcher while observing the two inspectors during the field study and is contained in the last two columns of Table 3 above. While the completion of the checklists using the scanner to record the data was twice as slow as the manual checking of the appropriate responses, the typed data entry of the manually completed checklist information into the Excel spreadsheet took a total of 153 minutes, or 5.46 minutes on average per inspection checklist. This is compared to the less than two minutes to upload all 28 scanned checklists data into a text file which was then imported into an Excel spreadsheet. This was a data entry time reduction of approximately 150 minutes. Thus, in this field study the scanner method was 65.2 % faster than the manual method in providing the data necessary to produce reports and evaluate possible trends in quality.


Summary

Identified benefits derived from bar coding of the inspection checklist responses could include, among others, the speed and accuracy of information processing (Cooper, 1989; Fales, 1990; Marsh, 1997; Malovany, 1998). It can be concluded from the statistical analysis in this study that the bar coded checklist, in conjunction with the use of the PT40 scanner, did provide more accurate data collection. However, the data collection process using the PT40 scanner was not perfect. The operator of the scanner can still make data collection errors by missing questions, scanning the incorrect answer to the questions and not correcting wrong answers before proceeding to the next question. The data collector must consciously concentrate on the data collection process and read the screen prompts on the PT40 to ensure the collection of the correct inspection data.

The scanner method produced reports that were 98.8% accurate compared to 60% for the manual method, and the scanner method was 65% faster in providing the data necessary to produce inspection reports and evaluate possible trends in quality. These study results indicate that the process developed by the researcher is faster and more accurate than the traditional manual data collection method and report generation. Thus, adoption of ADC technology can increase the amount of accurate and timely quality inspection data for residential construction companies. This could be expanded upon by using wireless technology and scanner units that also function as PDA’s with screens that can accept written notes as well as keyed and scanned data entry.


References

Albright, B. (2001, March). Construction industry adds mobile computing to toolbox. Frontline Solutions, 2(3), 52-53.

Becker, N. (1993). The complete book of home inspection (2^nd ed.). McGraw-Hill, Inc.

Cauldwell, R. (2001). Inspecting a house. Newtown, CT: The Taunton Press.

Cooper, T. (1989). Construction related applications of barcode technology. Proceedings ASC Annual Conference, 25, 27-32. University of Nebraska- Lincoln.

Fales, J. (1990, September). Who cares about data collection? Industrial Engineering, 22(9), 16-17.

Finch, E., Flanagan, R., & Marsh, L. (1996). ADC application in construction. Construction Management and Economics, 14, 121-129.

Hoffman, G. (1985). How to inspect a house: Exactly what to look for before you buy. Addison-Wesley Publishing Company.

Howard, H. (1987). The home inspection handbook. Garden City, NY: Doubleday & Company, Inc.

Irwin, R. (1995). The home inspection troubleshooter. Dearborn Financial Publishing, Inc.

Marsh, L. (1997). Bar coding for construction. Construction Computing, 54, 30.

Malovany, S. (1998, June). Bar coding: A benchmark of efficiency. Foodservice Equipment & Supplies, 51 (7), 65.

McNeill, J. (1979). Principles of home inspection. New York: Van Nostrand Reinhold Company.

Rogers, L., & Christofferson, J. (1997). Standardizing residential construction procedures: NAHB production manual. Proceedings ASC Annual Conference, 33, 195-202.

Rogers, L., & Christofferson, J. (2002). Building quality: An operations manual for home builders. NAHB Business Management & Information Technology Committee; Single Family Production Builders Committee.

Sawyer, R. (2003, June 23). Taking job data to the air: Cell phone technology and wireless networks pave the way. ENR, 250(24), 26.

Sheskin, D. (1997). Handbook of parametric and nonparametric statistical procedures. Boca Raton, FL: CRC Press LLC.

Staff (2002, July). Just like paper. BuilderOnline Magazine [WWW document]. URL http://builderonline.com/article-builder.asp?channelID=59&ArticleType=1&ArticleID=100

Torbica, Z., & Stroh, R. (1999). An assessment model for quality performance control in residential construction. Proceedings ASC Annual Conference, 35, 363-370. California Polytechnic State University, San Luis Obispo, California.

Traister, J. (1997). Home inspection handbook. Carlsbad, CA: Craftsman Book Company.

Appendix A

Figure 1. Partial Bar coded framing inspection form.

Figure 2. Inspection program logic diagram.


Appendix B

Figure 3. Figure 1-1 of PSC Falcon PT40 Product Reference Guide.


PSC PT40 Operating Instructions
Four main light blue Function Keys are located at the top of unit.

Under those are CLR which clears any current typed information before Entered,
SCAN to operate the scanner laser beam and ENT for Enter.

Below those buttons are the number keys which can also be used for typing letters if the red SH, or Shift, button is pushed. Push SH once to go from numbers to letters, and again to return to numbers.

Diamond shape arrow or Navigation Keys can be used to show information on the screen that may be off screen in any direction.

The green circular button is the Power Key.

Next to the Power button is a sun figure to indicate a Backlight Key for the view screen. Push once for the light to come on, again for the light to turn off.

Starting a Scanned Inspection

Push the Green power button.
Press the F3 function button.
If the Date displayed is correct, push the ENT button for Enter. If the date is not correct, enter the correct date in month, day, and year and then hit the ENT button for Enter. If an incorrect number is typed hit the CLR or Clear button to go backwards in order to correct the typed numbers- this must be done BEFORE hitting the ENT button.
If the time displayed on the screen is correct hit the ENT button. If the time is not correct, enter the correct time in hours, minutes, and seconds and then hit the ENT button for Enter. If an incorrect number is typed hit the CLR or Clear button to go backwards in order to correct the typed numbers- this must be done BEFORE hitting the ENT button.
Scan the Community where the inspection is to be done from the bar code list provided.
If the Community name displayed is correct hit 1 then the ENT button for Enter. If the Community name is not correct, hit the F4 function button to re-scan the correct Community name.
Type on the number keys the Lot or House number of the inspection location then hit the ENT button.
If the Lot# displayed on the screen is correct hit 1 then ENT. If the Lot# is incorrect hit the F4 function button to re-enter the correct Lot#.
Scan the appropriate Project Manager name from the bar code list provided.
If the Project Manager name displayed is correct hit 1 then ENT. If the Project Manager name is not correct hit the F4 function button and then re-scan the correct Project Manager name.
Scan the appropriate Subcontractor name from the bar coded list provided.
If the Subcontractor name displayed is correct hit 1 then ENT. If the Subcontractor name is not correct hit the F4 function button and then re-scan the correct Subcontractor name.
Scan the bar code for the appropriate question answers as indicated on the bar coded inspection form.
If the Question answer displayed is correct hit 1 then ENT. If the answer is not correct hit the F4 function key to re-scan the correct answer.
If the appropriate answer is No- then you have the option of scanning a Comment Form number where details of the problem can be explained. Peel the Comment Form number off the label form provided and paste this number on the last sheet of the Checklist.
If a Variance or Back-charge is needed scan a Variance number from the list provided. Peel the Variance form bar code off the label form provided and paste it on the last page of the Checklist. If not then hit the F4 function key to go to the next inspection question.
The last inspection question answer will be displayed. Make sure to answer the next question on the inspection form in numerical order.
Continue until all inspection questions are answered. Hit the F4 function key button to end the inspection.


To begin another inspection start over at instruction #1.