New information as of 23-Oct-2021

• LC_ and LT_ functions now determine both the standard error similar to how predict() works
  as well as the covariance of the model to be used by ratio_test()
• ratio_test() was discovered to be inaccurately calculating test statistic and p value
  when log_x = FALSE. This has been corrected when determining t by taking the log of each doses
• citation for the package has been updated as the paper the package was writing for has now been published
• v1.4.4 will be pushed to CRAN within a few days

Changes in function features

• LC_ and LT_ functions now have warnings messages for not supplying a variable for the weights argument. This is due to only needing to supply if you are taking the response / total for your response variable within the formula call of LC_probit. Otherwise if you use cbind(response, non-response) method you do not need to supply weights. If you do the model will be incorrect. If you don't supply weights there is a warning that will help you to make sure you are using one method or the other.

• This stackExchange question explains the differences to using cbind() vs. respone / total method. cbind() function in R for a logistic regression

• Test coverage is now 100% for all functions including ratio_test()
• LC_ and LT_ functions now have warnings for not supplying a value for p
• LC_ and LT_ functions now have error messages for not supplying a variable for the weights argument.
• ecotox no longer relys on ggplot2 as ggplot2 was a dependency as the examples for LC_ and LT_ functions use ggplot2 to plot the results. Instead these examples have now been commented. To run this part of the example just uncomment the lines. This was done as ggplot2 does not need to be a dependency for ecotox. tibble and stats are now the only dependency with stats being the only truely necessary dependency for ecotox. tibble is being used instead of R's default dataframe as tibbles are more useful but not necssary for the anlysis

• New function, ratio_test. This function allows the user to caculate differences between two seperate LC or LT values using a ratio test derived from Wheeler et al. 2006. To use the function, the user needs to create two
  glm() objects with either a probit or logit link. Model objects then can be passed to ratio_test() along with a percent argument which is the same argument as the p argument in LC_ and LT_ functions. A type argument is also needed to specify the link function used in the glm(). This can be either "probit" or "logit".
• All functions now output a tibble instead of dataframe
• New argument log_base for all LC, LT, and ratio_test functions. It
  allows the user to specifcy the base used when log transfoming the dose/x variable.
• All functions now allow for an LC or LT value to be calculated down to the
  1.0e-16 of a percent however four decmial places is typically appropriate (e.g. LC75.5). Tibble output will usually round the p value to the nearest whole number but if you insepect the p column or export to excel using openxlsx package you'll see p in the output is the same value supplied in the function call.

New Feature as of 20-Sep-2018

• All functions now have long_output argument which allows the user to
  decide whether or not a short or long output is displayed. Default is TRUE which will display long output of all 19 variables. If FALSE is used only 7 variables will be displayed.

New Feature as of 28-June-18

• All functions now have log_x argument which allows the user to specify if
  the x variable was log10 transformed or not. Default is TRUE which will
  output back transformed results. If FALSE results will not be back
  transformed.

New Feature as of 19-Mar-2018

• v1.3.1 is now available on CRAN.
• development of ratio_test is underway and will be implemented soon
• development of spearman_karber is still occurring
• shiny app is being development

New Feature as of 9-Feb-2018

• All functions now have subset argument
• Data argument can now be just the dataframe of interest
  no need to use brackets to subset
• Weights is no longer an optional argument and can be specified
  just by object name. Again no need to add name of dataframe,
  bracket to subset [] and $ to select object used for weights.
• LC_probit and LC_logit and LT_probit and LT_logit might be
  depreciated as lc_probit and lc_logit and lt_probit and lt_logit
  are simpler function calls for users and are consistent with
  naming conventions. Thoughts??

New functions which allow the user to run a logit analysis. LC and LT have been depreciated and will be removed in future versions.