# Difference Measurement (compare means) between two continuous variables

#### yologer

##### New Member
Hello,

I'm looking for a test and need some help ..

Variable 1: value (continuous)
Variable 2: expectation (continuous)
Dependent variable: performance

My hypothesis: high value and low expectation lead to lower performance than low value and low expectation.

How do I test this? I thought to use a t-Test, but I have no clue how to split my variable into high and low? My next thought was to use simple slope, but there are no statistical prove (only graphically).

Is there any solution?

Thank you all

#### hlsmith

##### Less is more. Stay pure. Stay poor.
How is "performance" formatted?

#### yologer

##### New Member
Hello @hlsmith, thank you for your response!

My dependend variable "performance" is also a continuous variable.

#### hlsmith

##### Less is more. Stay pure. Stay poor.
Are you familiar with multiple linear regression? What is the sample size? Have you look at they variables on a 3 dimensional graph?

The easiest option is to run a regression model with Var1 and Var2 predicting performance. In the model output you will be able to see a positive or negative relationship with the variables and outcome.

To test if high or low groupings predict differently, then you may be looking at quantile regression.

#### yologer

##### New Member
Hello @hlsmith, thx for the response!

I've never heard of quantaile regression (and I think, it's not integrated in SPSS, which I have to use).

My first solution is to run a moderation analysis with PROCESS and to look at the standard deviation (+/- - high/low) - is this a proper way?

"The easiest option is to run a regression model with Var1 and Var2 predicting performance. In the model output you will be able to see a positive or negative relationship with the variables and outcome."

I think it's not enough.

I have to detect if:
low expectation / high value = high performance
low expectation / low value = low performance

#### hlsmith

##### Less is more. Stay pure. Stay poor.
How do others define high and low. If you create your own evidence based definition it may not be generalizable to other samples. Perhaps you can use standardization as a way to split data, it they look normally distributed.