Print Page   |   Contact Us   |   Report Abuse   |   Sign In   |   Register
TL v56n2 Staffing Needs of the Reference Desk: A Statistical Approach
Share |
 

 

Tennessee Libraries

Volume 56 Number 2

 2006

 

Staffing Needs of the Reference Desk

at the University of Tennessee at Chattanooga:

A Statistical Approach 

by

Sarla Murgai
Reference and Instructional Services
University of Tennessee at Chattanooga

Power Point Presentation (in .pdf format)

Summary: This study grew out of a need to assess reference desk data and determine: 1) the number of questions handled for the last five years; and, 2) the times when the desk should be double staffed. Analysis of the data by chi-square standardized residuals identified the days of the week, and the times during the day of heavy and light use. Data sort identified the heavy/light use weeks, months, and semesters. The variability and the range of data were also established through descriptive analysis. When data were compared from week to week for the random sample years with the year when the data were collected by every hour for every day of the week, a very similar pattern was revealed. Heavy days, weeks, hours and months fell into a similar pattern from semester to semester and from year to year. Other academic libraries can follow this model and apply it to their work environment after adjusting for their academic calendar and user behavior.


 

This research study analyzes reference desk statistics collected by the hour for all of the 2001-2002 academic year and compares it with the 2002 to 2006 data collected by the sampling method. An Excel spreadsheet was used to enter the data by day and hours of the week for 53 weeks from August 2001 though July 2002.

Table 1. The number of questions handled for the year 2001-2002, hourly.

The data in Table 1 were analyzed using standardized residuals from a chi-square test. Chi-Square test is a test for “goodness of fit” between observed values (O) as computed and expected values (E) (Flat or uniform values, usually the mean). A statistical test calculated as a sum of squares of observed values minus the expected values divided by the expected values. (O-E) 2 /E (see Appendix for statistical definitions)

Table 2. A selected sample of observed and expected values by the hour of the day.

Number

Monday
7:50-9

Monday
9-10

Monday
10-11

Monday
11-12

Monday
12-1

Monday
1-2

Monday
2-3

Observed

79 day/hrs

100

175

201

239

282

249

220

Expected

 

181.36

181.36

181.36

181.36

181.36

181.36

181.36

O-E

 

-81.36

-6.36

19.64

57.64

100.64

67.64

38.64

(O-E) 2

 

6619.4496

40.4496

385.7296

3322.3696

10128.4096

4575.1696

1493.0496

(O-E)2/E

 

36.49895015

0.223034848 

2.126872519

18.31919718

55.84698721

25.2270049

8.23251875

Residuals

 

-6.0414361

-0.47226565

1.458380101

4.280093127

7.473084183

5.02264919

2.86923661

Table 3. Chi-Square Standardized Residuals

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

7:50-9

-6.04144

-5.81867

-6.41271

-3.81377

-6.11569

 

 

9-10am

-0.47227

-0.02673

2.052425

-3.07121

-0.10099

 

 

10-11

1.45838

4.057326

6.507761

1.978169

2.275192

-5.52165

 

11-12

4.280093

4.280093

8.735429

3.463281

6.136483

-4.70484

 

12-1

7.473084

6.953295

8.883941

3.017748

1.012846

-5.44739

 

1-2

5.022649

4.50286

8.512663

2.349447

0.715824

-6.11569

 

2-3

2.869237

1.829658

6.730528

2.052425

-1.43759

-3.29398

5.17116

3-4

3.983071

3.834559

2.943492

2.275192

-3.73951

-3.66526

0.938591

4-5

4.131582

3.686048

1.978169

-2.18014

-2.77419

 

2.497959

5-6

1.161358

-0.10099

-0.54652

-1.36333

 

 

 

6-7

0.864335

1.606891

-4.2593

-5.07611

 

 

-3.66526

7-8

0.715824

2.720725

-2.18014

-4.55632

 

 

-5.96718

8-9

-0.69503

-1.36333

-4.9276

-6.78399

 

 

-8.2691

9-10pm

-1.73461

-2.47717

-5.96718

-7.15527

 

 

-13.467

When the value of residuals is greater than or equal to 2, the cell is a major contributor to the significance (Dennison, 160). As the value of the cell grows larger, so does the contribution of the cell to the significance. Negative values are as significant as the positive values. The direction of the value simply shows if the major contributor deviates significantly above or below zero. The numeric value shows how great the deviation is. The negative values indicate reduced staff needs as much as the positive values indicate increase in staffing needs.

This analysis showed variation in reference usage as is evident from the bolded highlighted numbers. Of the 79 hours and 50 minutes examined, 29 hours have a value of 2 or above and are thus positive contributors to the reference desk load . The negative numbers are as significant as the positive numbers, but they are not bolded because they are not yet being used to adjust staffing at UTC. Generally, the early morning hour of each day and late hours on Thursday through Saturday have significantly less reference usage. In general, significantly greater reference usage occurs on Monday from 11:00 a.m. to 5:00 p.m.; Tuesday from 10:00 a.m. to 1:00 p.m. and 3:00 p.m. to 5:00 p.m.; Wednesday for 9:00 a.m. to 4:00 p.m.; Thursday from 11:00 a.m. to 4:00 p.m.; Friday 10:00 a.m. to 12:00 p.m.; and Sunday from 2:00 p.m. to 3:00 p.m. and 4:00 p.m. to 5:00 p.m., as indicated by the bold numbers.

Figure 1. High and the low usage times of the day during the week for 2001-2002.

 

Figure 2. Total number of weekly questions during the year 2001-2002

Figure 3. Number of weekly questions arranged in a descending order for the year 2001-2002

Figure 4. Number of questions by time of day and day of the week, Monday-Friday 2001-2002.

Figure 5. Percentage of questions received, by day of the week.

Table 4. Total number of questions handled during the five years of study by the month for 2001-2006

2001-2002 2002-2003 2003-2004 2004-2005 2005-2006
Sep 1781 2164 2502 2116 1710
Oct 1805 1633 1893 1854 1598
Nov 1501 1782 1725 1386 1233
Dec 404 345 763 506 434
Jan 1352 1666 1444 1202  
Feb 1364 1358 1351 1206  
March 1419 1620 1444 929  
April 1464 1355 1096 1134  
May 728 516 632 554  
June 645 621 600 652  
July 616 608 682 752 617
Aug 1100 917 1622 920 527

Figure 6. Reference questions by month: 2001-2005.

Figure 6 shows the similarity in the number of questions by months for four years. The summer months and December are typically low in the number of questions posed at the reference desk. During the spring and the fall semesters the number of questions is much higher. It clearly shows that other academic libraries could adopt this method for analyzing their reference load during the year.

During 2003-2004 Lupton Library decided to change the policy. Starting with September 2002 the reference questions were divided into two categories Internal (coming from patrons in the library) and External (coming from patrons outside the library by phone). Each of these two categories were further divided into three subdivisions: 1) directional questions; 2) questions lasting less than five minutes; and, 3) questions lasting more than five minutes. Standard definitions were also added to the three types of questions. Tables 5 and 6 show the pattern of the distribution of those questions.

 

 

Table 5

Reference Desk Statistics 2003-2004

Internal Users

External Users

Directional

Less than
5 minutes

More than
5 Minutes

Directional

Less than
5 minutes

More than
5 Minutes

E-mail
Questions

Monthly
Totals

Month

July

48

411

67

25

115

12

4

682

August

232

1126

80

31

116

17

20

1622

September

248

1844

150

32

184

24

20

2502

October

181

1354

153

31

134

16

24

1893

November

143

1260

113

45

145

10

9

1725

December

121

440

44

50

94

5

9

763

January

209

960

116

33

102

14

10

1444

February

172

872

157

37

83

11

19

1351

March

65

976

211

21

140

10

21

1444

April

100

665

177

28

83

27

16

1096

May

62

363

79

31

76

7

14

632

June

82

312

88

30

76

8

4

600

 

 

 

 

 

 

 

 

 

Yearly
Totals

1663

10583

1435

394

1348

161

170

15754

Table 5 presents the internal and external questions; and under each category the directional, less than 5 minutes and more than 5 minute questions.

 

Table 6

Reference Desk Statistics 2002-2003

 

Internal Users

External Users

 

 

 

Directional

Less than
5 minutes

More than
5 Minutes

Directional

Less than
5 minutes

More than
5 Minutes

E-mail
Questions

Monthly
Totals

Month

 

 

 

 

 

 

 

 

July

 

 

 

 

 

 

4

 

August

 

 

 

 

 

 

13

 

September

211

1555

197

38

143

20

14

2178

October

137

1183

141

30

121

21

7

1640

November

139

1373

143

14

113

0

11

1793

December

5

410

30

30

40

0

11

526

January

127

1205

80

22

213

19

23

1689

February

104

927

137

18

168

4

17

1375

March

79

1158

124

23

232

4

17

1637

April

98

1019

98

14

112

14

14

1369

May

68

325

29

21

66

7

8

524

June

54

377

54

25

104

7

2

623

 

 

 

 

 

 

 

 

 

Yearly
Totals

1022

9532

1033

235

1312

96

141

13354

Note: The breakdown of reference questions into internal and external users and the subdivisions of data into directional, less than 5 minutes and greater than 5 minutes started in September of 2002 -2003.

Figure 7.

Figure 7 shows that the maximum number of questions are answered within 5 minutes (10,553) followed by internal directional questions (1663); closely followed by internal >5 minutes questions (1435); followed by external < 5 minutes (1348). Figures 8 and 9 show the percentage of questions in each of those categories.

Figure 8. Percentage of Reference Questions 2003-2004

 

Figure 9. Percentage of Reference Questions 2002-2003  

 

The statistics for 2002-2003 show that the greatest numbers (66% to 70%) of reference questions come from internal users and take less than five minutes to answer. The directional questions range from 8% to 11% for internal users and 2% to 3% from external users.

All those questions that take more than five minutes however, involve orienting the student to the library catalog, and/or to the databases to which library subscribes, or has links to, and other print or online sources. It involves teaching him/her as to how to use the same sources efficiently. For some users who are not computer savvy, it may even take longer and may take one or two such repeated sessions. I am reminded of a remark that one of the librarians made on the statistics sheet “It took forever to answer that one question.” The main goal is to get more help with answering questions at the reference desk so that the librarians do not feel rushed in performing their duty. Some of these reference questions a professional librarian carries home with him/her, long after the patron has left, as Dr. S.R. Ranganathan (1961, Introduction) would say, “A reference librarian sleeps and eats with reference questions.”

Figures 10 to 14 show that patterns of workload are similar during the days of the week throughout the study. One explanation for unusual dips in the graphs is that sometimes reference librarians get so busy that in the middle of a busy hour they forget to mark the questions.

Figure 10. 3rd Week of September, Monday

 

Figure 11. 1st Week of September, Tuesday

 

Figure 12. 2nd Week of November, Wednesday

 

Figure 13. 2nd Week of February, Thursday

Figure 14. 1st Week of September, Friday

 

Conclusions: Applying the data analysis

At Lupton Library, we found that the heaviest use times were Monday from 11:00 a.m. to 5:00 p.m.; Tuesday from 10:00 a.m. to 1:00 p.m. and 3:00 p.m. to 5:00 p.m.; Wednesday for 9:00 a.m. to 4:00 p.m.; Thursday from 11:00 a.m. to 4:00 p.m.; Friday 10:00 a.m. to 12:00 p.m.; and Sunday from 2:00 p.m. to 3:00 p.m. and 4:00 p.m. to 5:00 p.m. Usage was lowest during the early morning hour of each day and late hours on Thursday through Saturday.

As a result of these findings, the Reference department at the Lupton Library implemented double staffing of the reference desk from 11 a.m. to 1 p.m. Monday through Thursday starting with the fall semester of September 2002. During spring 2005, due to higher demand on instruction and also because two librarians had left the department, librarians from other departments were pooled to fill in during 11 a.m. to 1 p.m. for the double staffing hours.

Staffing decisions were not made strictly according to the findings of this study, because availability of the staff, instruction load, and convenience of scheduling, were also taken into account. For example, since there is only one staff member on duty during weekends and nights, staffing cannot be further reduced without cutting hours of service. Similarly, even though on Tuesdays from 7 p.m. to 8 p.m., and on Sundays from 2 p.m. to 3 p.m. and 4 p.m. to 5 p.m., the load indicates a need for a second librarian, there is only one librarian on duty. This individual handles the problem of queuing, if it occurs. Most of the discussion in the literature deals with the number of questions asked, the time lag between them, and the need to better utilize the professional librarian's time. This method identifies the high and low use times in this library. A well trained paraprofessional could help during the time when the demand is less.

Even though reference is essentially a human science, efforts are underway to automate and standardize as many aspects of this service as possible. In order to improve the efficiency and productivity of the reference services it is essential to measure various aspects of reference process. This study suggests that statistical analyses of reference transaction count can be used as a method of documenting how busy library is at a particular time, since heavy days, weeks, hours, and months fall into similar patterns from semester to semester and year to year. Managers can use this information to make sound decisions about staffing needs at the reference desk.

References

Dennison, R. F. 1999. Usage-based staffing of the reference desk: a statistical approach. Reference and User Services Quarterly 39(2), 158-65.

Ranganathan, S. R. 1961. Reference Service . London : Asia Publishing House.

Additional Resources

Blandy, S. G., Martin, L. M. and Strife, M. L., eds. 1992. Assessment and accountability in Reference work. Binghamton , N.Y. : Haworth Press.

Bundt, J. 2000. Strategic stewards: Managing accountability, building trust. Journal of Public Administration Research & Theory, 10(4), 757-778.

Chu, F. 1997. Another look at staffing the reference desk. College & Research Library News , 58(10), 713.

Coffman, S. & Saxton, M. 1999. Staffing the reference desk in the largely-digital library. The Reference Librarian , 66, 141-63.

Cummins, T. R. 1992. Developing personnel and staffing standards. Library Administration & Management. 6(4):182-186.

Ferguson, C. D. & Bunge, C. A. 1997. The shape of services to come: Values-based reference service for largely digital library. College & Research Libraries 58(3), 252-265.

Glockner, B. 2004. Accountability and accreditation for special libraries: It can be done! The Australian Library Journal , 53(3), 277-284.

Kantor, P. B. 1984. Objective performance measures for academic and research libraries . Washington , D.C. Association of Research Libraries.

Kesselman, M. and Watstein, S. B. 1987. The measurement of Reference and information services. The Journal of Academic Librarianship 13(1), 24-30.

Kuhlman, J. R. 1995. On the economics of reference service: Toward a heuristic model for an uncertain world. The Reference Librarian 49/50, 25-43.

Lochstet, G. and Lehman, D. H. 1999. A correlation method for collecting reference statistics. College and Research Libraries 60(1), 45-53.

Maxstadt, J. M. 1988. A new approach to reference statistics . College & Research Libraries News 49(2), 85-86, 88.

Murgai, S. R. 1986. Future challenges in library science. The Georgia Librarian , 23(3), 68-69.

Rettig, J. 2004. Interview on the future of the academic library. Retrieved July 20, 2005 , from http://faculty.jscc.edu/scohen/rettiginterview.html.

Richardson, J. V., Jr. 2002. Reference is better than we thought. Library Journal , 127(7), 41-42.

Schlesinger, A. 1997. Staffing pattern questions at the Cleveland Heights-University Heights public library . (Report No. IR 057 232). Cleveland , OH : A Master's Research Paper submitted to the Kent State University School of Library Science. (ERIC Document Reproduction Service No. ED424867).

Summerhill, K. S. 1994. The high cost of reference: The need to reassess services and service delivery. The Reference Librarian , 43, 71-85.

Taylor, A. 1994. Plan for service: Professional and non-professional reference staff. The Reference Librarian 43, 101-105.

Witucke, V. and Schumaker, C. J. 1991. Analyzing reference activities: The affordable solution. RQ 31(1), 58-69.


Appendix

Definitions of Statistical Terms

Chi-Square test: is a test for “goodness of the fit” between observed values (O) as computed and expected values (E) (Flat or uniform values, usually the mean). A statistical test calculated as a sum of squares of observed values minus the expected values divided by the expected values. (O-E) 2 /E

Uniform Distribution: The probability distribution of a random variable having constant probability over an interval.

Standard error: The standard of deviation of the errors of sampling distribution of a statistic.

R = Residual: The difference between the actual value of a single measurement minus the computed value regarded as the most probable value is often the arithmetic mean of a number of similar measurements, The residual then being a deviation from the mean. It is also called residual error and error of estimate. The residuals are ordinarily squared,

Null Hypotheses: The “no difference” or “association” hypotheses to be tested against an alternative hypothesis.

 


Membership Software Powered by YourMembership  ::  Legal