+ Reply to Thread
Results 1 to 7 of 7

Thread: Linear regression / trend question - basic question

  1. #1
    Points: 4,373, Level: 42
    Level completed: 12%, Points required for next Level: 177

    Posts
    15
    Thanks
    4
    Thanked 0 Times in 0 Posts

    Linear regression / trend question - basic question




    Sorry for basic question:

    I have a series of measurements taken at weekly intervals for a period of 20 weeks. Each measurement is a proportion: i.e. there is a numerator and a denominator for each reading. The denominator varies from week to week.

    When I plot it on a scatter plot, there is a clear visual trend for the proportion to increase over time.

    I would like to test the statistical significance of the trend and am looking for advice about what kind of test to use. Any pointers would be helpful. Happy to share the data if that helps.

    Again, apologies for straightforward question.

    Thanks,
    T

  2. #2
    TS Contributor
    Points: 17,779, Level: 84
    Level completed: 86%, Points required for next Level: 71
    Karabiner's Avatar
    Location
    FC Schalke 04, Germany
    Posts
    2,542
    Thanks
    56
    Thanked 640 Times in 602 Posts

    Re: Linear regression / trend question - basic question

    Spearman rank correlation between "week" and "proportion".

    With kind regards

    Karabiner
    »Jetzt kann mich der Führer mal am Arsch lecken.« (Ernst Kuzorra, 1941)

  3. #3
    Omega Contributor
    Points: 38,396, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,001
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Linear regression / trend question - basic question

    If proportions typically land in the middle of 0-1 and not too close to "0" or "1", you may be able to use linear regression with week number being the predictor. If percentages are closer to the boundaries, beta regression may be an option.


    For simplicity, you could also fall back on Karabiner's suggestion as well.
    Stop cowardice, ban guns!

  4. #4
    TS Contributor
    Points: 14,811, Level: 78
    Level completed: 91%, Points required for next Level: 39
    Miner's Avatar
    Location
    Greater Milwaukee area
    Posts
    1,171
    Thanks
    34
    Thanked 405 Times in 363 Posts

    Re: Linear regression / trend question - basic question

    Another option is to use a p-chart, or even an individuals control chart. Changes in the underlying process are detected using rules. One popular set of rules are Nelson's Rules.

    These approaches are the standard approach in industrial statistics for dealing with data collected over time where the concern is with detecting and correcting changes in the underlying process.

  5. The Following 2 Users Say Thank You to Miner For This Useful Post:

    bugman (05-16-2017), noetsi (05-16-2017)

  6. #5
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Linear regression / trend question - basic question

    You could, if you think there is a trend over time, specify time as a predictor. Then see if its statistically significant. However, you would have to test for autocorrelation and if there is any run something like regression with autoregressive error. Otherwise your t test would not be valid since autocorrelation will effect your standard error.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  7. #6
    Super Moderator
    Points: 31,766, Level: 100
    Level completed: 0%, Points required for next Level: 0
    bugman's Avatar
    Posts
    2,255
    Thanks
    290
    Thanked 324 Times in 265 Posts

    Re: Linear regression / trend question - basic question

    Quote Originally Posted by Miner View Post
    Another option is to use a p-chart, or even an individuals control chart. Changes in the underlying process are detected using rules. One popular set of rules are Nelson's Rules.

    These approaches are the standard approach in industrial statistics for dealing with data collected over time where the concern is with detecting and correcting changes in the underlying process.
    Thanks for linking the Nelson Rules. This is going to be very useful. Although I have used control charts before, I hadn't heard of the Nelson Rules. Cheers mate.
    The earth is round: P<0.05

  8. #7
    Points: 4,373, Level: 42
    Level completed: 12%, Points required for next Level: 177

    Posts
    15
    Thanks
    4
    Thanked 0 Times in 0 Posts

    Re: Linear regression / trend question - basic question


    Hi. This is the OP. Very useful information. Thanks to all who posted.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats