Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

lisaz0224
Posts: 1
Joined: Fri Nov 07, 2014 9:39 pm

Hi everyone. I am a high school student conducting a research project. At this point
my data collection is complete and entered into SPSS. However, I am completely unfamiliar with the program and I have a lot of data, so I need heavy assistance with things many of you may consider very basic. I would be very grateful if you could explain to me step-by-step how to accomplish each of these tasks. Here is a summary of my project, and the data I have already entered into the program:

I have four lists of forty words each, named 1, 2, 3, and 4. In lists 1, 2, and 3 there are two corresponding variables for each word (let’s call them variables a and b). Therefore, each of those lists contains eighty variables. List 4 is a little different; it contains the same variables a and b as the first three lists, but it contains an additional variable ‘c’. I would like to attach a diagram, but here is the best I can explain. The columns across the top of the page go like this:

|word1a | word1b | word2a | word2b|...|word40b|word41a | word41b |...|word80b| word81a |...|word120b|word121a|word121b|word121c|word122a|word122b|word122c|...|word160c|.

Variables of List 1 are black and bold
Variables of List 2 are black
Variables of List 3 are red
Variables of List 4 are red and bold

I have 360 columns in my data sheet, and I have 50 rows. All the data is numerical (exclusively 1s and 0s), or missing. Here is what I would like to do:

1.) Assign call-tags (I believe they are called ‘attributes,’ but I do not know how to work it) to some of my variables. If, for example, I could tag the first eighty variables with the name ‘List 1’ and simply use that when I would like to, for example, count the number of 1s for the first eighty variables for all fifty cases in sum, data analysis becomes much simpler.

2.) In that same vein, possibly use the boolean command (‘Include if case satisfies condition:’) to:

a.) Choose all cases (rows) containing any variables of a certain type of name. For example, ‘V1, V2, V3… V40’. This is especially difficult because some of the rows are missing data for certain variables. For example, Row 6 may have an empty cell corresponding to ‘V40’. All rows, however, are filled for at least one of the variables.
b.) Choose all cases containing a certain string of letters, in order. For example, I’d like to choose all cases of variables containing “ify”.

3.) For all cases of a certain attribute, find the ratio of 1s: total number of responses (So, the sum of the values in the cells divided by the total number of cases, not including missing values).
a. For all cases of a certain attribute WITHIN EACH ROW, calculate the same ratio. Afterward, sort the rows in increasing order of accuracy.
b. Using that order of rows, create quintiles (10 rows each) and calculate the same value within each of the ordered quintile. For example, the top quintile would have a certain ratio, the second a different one, etc.
c. Do the same with all cases of TWO OR MORE different attributes.

4.) For each of multiple columns (in my project, columns of variable ‘c’ in List 4), look at two different columns (in my project, column Word121c in list 4 would call Word121b and Word81b)
a. If either of those two columns (column Word121b and Word81b) are empty, do nothing.
b. If the values in those two columns match, enter a ‘1’ as a value in the column .
c. If the values in those two columns do not match, enter a ‘0’ as a value in the column.

I think that’s pretty much it (a lot, I know). If you could help with ANY ONE of these issues, or even link me to something helpful, I would be eternally grateful to you.

Thank you so much!
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

lisaz0224 wrote: 1.) Assign call-tags (I believe they are called ‘attributes,’ but I do not know how to work it) to some of my variables. If, for example, I could tag the first eighty variables with the name ‘List 1’ and simply use that when I would like to, for example, count the number of 1s for the first eighty variables for all fifty cases in sum, data analysis becomes much simpler.
I am not familiar with attributes in spss.
However, in you case I would create variable names Like this:

List_1_01 List_1_02 .... List_1_30 etc.

Then you can use this in syntax: List_1_01 to List_1_30.
2.) In that same vein, possibly use the boolean command (‘Include if case satisfies condition:’) to:
look into if-functions in syntax.
a.) Choose all cases (rows) containing any variables of a certain type of name. For example, ‘V1, V2, V3… V40’. This is especially difficult because some of the rows are missing data for certain variables. For example, Row 6 may have an empty cell corresponding
to ‘V40’. All rows, however, are filled for at least one of the variables.
look into select cases.
about your missing data: Spss does missing is listwise by default for most analyses, so the case would not be included if there is a missing on at least one variable. You can change this however.
b.) Choose all cases containing a certain string of letters, in order. For example, I’d like to choose all cases of variables containing “ify”.
"choose cases" is usually done with the select cases option, but you may have to create a new variable which to select cases on. Also, what do you mean by "choose"? And what do you mean by "cases"? I mean rows.
If you want to choose cases of variables (I assume you don't mean the variable name, thus the entire column, but only the specific value for a specific case?) .
To find a string in a cell, look into "index".
3.) For all cases of a certain attribute, find the ratio of 1s: total number of responses (So, the sum of the values in the cells divided by the total number of cases, not including missing values).

a. For all cases of a certain attribute WITHIN EACH ROW, calculate the same ratio. Afterward, sort the rows in increasing order of accuracy.
this depends on whether you want to do it by row or by column.
but look into:
- aggregate (find mean for column)
- compute (e.g.:
compute newvar = var1 / (var1 + var2 + var3).
execute.
b. Using that order of rows, create quintiles (10 rows each) and calculate the same value within each of the ordered quintile. For example, the top quintile would have a certain ratio, the second a different one, etc.

c. Do the same with all cases of TWO OR MORE different attributes.
I don't get this, but you could probably use sort cases - create ID variable for this.
4.) For each of multiple columns (in my project, columns of variable ‘c’ in List 4), look at two different columns (in my project, column Word121c in list 4 would call Word121b and Word81b)
a. If either of those two columns (column Word121b and Word81b) are empty, do nothing.
b. If the values in those two columns match, enter a ‘1’ as a value in the column .
c. If the values in those two columns do not match, enter a ‘0’ as a value in the column.
this could all be achieved via if-functions, so look that up.
I think that’s pretty much it (a lot, I know). If you could help with ANY ONE of these issues, or even link me to something helpful, I would be eternally grateful to you.

Thank you so much!
why would you want to use spss for this? It sounds like you only want to create new variables, and not really do any type of inferential statistics.
For someone who uses spss a lot, it is probably not too difficult to do this using spss, but if you are not used to spss, why not use a more accessible program like excel? This is more intuative, for most people.

Who is online

Users browsing this forum: No registered users and 1 guest