A typology and evaluation of the survey sample designs in the Educational Administration Quarterly: 1980-1995 /

This inquiry analyzed 53 sample designs from 50 survey

Bibliographic Details
Main Author: Wang, Lin
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1996.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=739327361&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:This inquiry analyzed 53 sample designs from 50 survey
articles published in
the Educational (EAQ) from 1980 through 1995. Specifically,
the inquiry developed a typology of sample selection methods,
and evaluated the quality of the sample designs. Sampling
Designs. Three typological models, the one-stage, two-stage,
and three-stage models, were developed from the data with the
following findings: 1.The one-stage and two-stage models
contained 29 and 21 of the 53 sample designs, respectively.
2.Stratified random sampling was used in 10 of the 18 one-
stage probability samples, whereas criterion-based selections
occurred in 9 of the 11, nonprobability samples. 3. A
typical two-stage sample involved nonprobability selections
at stage one, followed by simple random sampling at stage
two. Selection Procedures. The evaluation of the sample
selections found that: 1. The target or survey population was
not explicitly defined in 42 designs, and the unit of
analysis was not mentioned in 30 designs. 2. Sample size
decisions were mentioned in only two designs No information
was given about sample size allocation methods for any of the
stratified samples. 3. Response rates were reported in 49
designs. These response rates ranged from 34% to 100%.
Estimation Procedures. Three major findings resulted from
evaluating the estimation procedures: 1.In the designs
that must have involved unequal selection probabilities, the
required weighting adjustments were not performed. 2.
Sixteen probability samples used complex designs. No
required adjustments were reported in variance estimation.
This problem results in underestimated variances and
nonexistent statistical significance for the cluster samples.
3. Not a single study specified a confidence interval for a
point estimate. Data Analysis. The evaluation of the data
analysis procedures yielded these three major findings: 1.
No adjustments for inflated error rates were mentioned
in the studies performing many statistical tests on the same
data. 2.Inferential statistics was applied to nonrandom
data without discussing implications for the interpretation
and generalization of the findings. 3.Nonresponse bias
effects were not investigated in any designs with low
response rates. The study concluded with research
recommendations and implications for the EAQ editors.
Item Description:Vita.
"Major Subject: Educational Psychology".
Physical Description:xiii, 255 leaves ; 28 cm.
Issued also on microfiche from University Microfilms Inc.
Bibliography:Includes bibliographical references: pages 228-237.