Wednesday, 28 September 2016

LONGITUDINAL STUDIES




In the mass media area, the first major longitudinal study was done by Lazarsfeld, Berelson, and Gaudet (1944) during the 1940 presidential election.
TYPES OF LONGITUDINAL STUDIES
The three main types of longitudinal studies are trend study, cohort analysis, and panel study. Each is discussed in this section.
Trend Studies
The trend study is probably the most com­mon type of longitudinal study in mass media research. Recall that a trend study samples different groups of people at differ­ent times from the same population.

Trend studies are useful, but they have limitations. Suppose that a sample of adults is selected three months before an election and 57% report that they intend to vote for Candidate A and 43% for Candidate B. A month later, a different sample drawn from the same population shows a change: 55% report that they are going to vote for A and 45% for B.



This is a simple example of a trend study

3 month before election
one month before
Before one week
A
57%
55%
45%
B
43%
45%
55%
 To determine both the gross change and the net change, a panel study is necessary.
Advantages.
·         Trend studies are valuable in describing long-term changes in a popu­lation.
·         They can establish a pattern over time to detect shifts and changes in some event.
·         They can be based on a comparison of survey data originally constructed for other purposes.


Disadvantages.
If data are unre­liable, false trends will show up in the results. Trend analy­sis must be based on consistent measures.
Examples of Trend Studies.
the trend study  about newspaper reading and attitudes to­ward ethnic minorities that spanned five years.
Cohort Analysis
 Cohort analysis attempts to identify a cohort effect: Changes in the dependent variable due to aging, or are they present because the sample members belong to the same cohort?

To illustrate, suppose that 50% of college seniors report that they regularly read news magazines, whereas only 10% of college freshmen in the same survey give this answer. How might the difference be accounted for? One explanation is that freshmen change their reading habits as they progress through college. Each survey has different participants—the same people are not questioned again, as in a panel study—but each sample represents the same group of people at different points in their college career.

Reading habit
Seniors
Freshers
2014
68
32
2015
55
45
2016
60
40
.
Typically, a cohort analysis involves data from more than one cohort. It displays news magazine readership for a number of birth cohorts. Note that the column variable (read down) is age, and the row variable (read across) is the year of data collection. This type of table allows a researcher to make three types of comparisons. First, reading down a single column is analogous.

Percentage of Adults Who Regularly Read News Magazines
Age
1992
1996
2006

18-21
15
12
10
22-25
34
32
28
26-29
48
44
35
A "pure" period effect. There is no variation by age at any period; the columns are identical, and the varia­tions from one period to the next are identi­cal. Furthermore, the change in each cohort (read diagonally to the right) is the same as the average change in the total population.


Table 8.2
Cohort Table Showing Pure
Age Effect




Year
Age
1992
1996        2000
18-21
15
10
22-25
20
15
26-29
25
20
Average
20
15

Second, reading across the rows shows trends at each age level that occur when cohorts replace one another. Second, influences associated with members in a certain birth cohort are called cohort effects.

It  shows a "pure" cohort effect. Here the cohort diagonals are constant, and the variation from younger to older respon­dents is in the opposite direction from the variation from earlier to later survey peri­ods. In this table, the key variable seems to be date of birth. Among those who were born between 1971 and 1974, news maga­zine readership was 15% regardless of their age or when they were surveyed.


Third, reading diagonally toward the right reveals changes in a single cohort from one time to another (an intracohort study). Finally, in­fluences associated with each particular time period are called period effects.

18-21
15
10
5
22-25
20
15
10
26-29
25
20
15
Average
20
15
10



Advantages.
1.      Cohort analysis is an ap­pealing and useful technique because it is highly flexible.
2.      It provides insight into the effects of maturation and social, cultural, and political change.
3.      A cohort analysis can be less expensive than experiments or surveys.
Disadvantages.
 The major disadvantage of cohort analysis is that the specific effects of age, cohort, and period are difficult to un­tangle through purely statistical analysis of a standard cohort table.
1.      In survey data, much of the variation in percentages among cells is due to sampling variability. There are no uni­formly accepted tests of significance appro­priate to a cohort table that allow researchers to estimate the probability that the observed differences are due to chance.

2.      A second disadvantage of the technique is sample mortality. If a long period is involved or if the specific sample group is difficult to reach, the researcher may have some empty cells in the cohort table or some that contain too few members for meaningful analysis.
Examples of Cohort Analysis. Cohort analy­sis is useful in the study of public opinion.


Advertisement